Face Generation¶

In this project, you'll use generative adversarial networks to generate new images of faces.

Get the Data¶

You'll be using two datasets in this project:

MNIST
CelebA

Since the celebA dataset is complex and you're doing GANs in a project for the first time, we want you to test your neural network on MNIST before CelebA. Running the GANs on MNIST will allow you to see how well your model trains sooner.

If you're using FloydHub, set data_dir to "/input" and use the FloydHub data ID "R5KrjnANiKVhLWAkpXhNBe".

data_dir = './data'

# FloydHub - Use with data ID "R5KrjnANiKVhLWAkpXhNBe"
#data_dir = '/input'


"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import helper

helper.download_extract('mnist', data_dir)
helper.download_extract('celeba', data_dir)

Found mnist Data
Found celeba Data

Explore the Data¶

MNIST¶

As you're aware, the MNIST dataset contains images of handwritten digits. You can view the first number of examples by changing show_n_images.

show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
%matplotlib inline
import os
from glob import glob
from matplotlib import pyplot


# import numpy
import numpy as np

mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'mnist/*.jpg'))[:show_n_images], 28, 28, 'L')


# Had to convert the image to np.array
# This solution was given to me by 'Itai Dror' in the Slack Chat

pyplot.imshow(np.array(helper.images_square_grid(mnist_images, 'L')), cmap='gray')

# pyplot.imshow(helper.images_square_grid(mnist_images, 'L'), cmap='gray')

<matplotlib.image.AxesImage at 0x249859d5d68>

CelebA¶

The CelebFaces Attributes Dataset (CelebA) dataset contains over 200,000 celebrity images with annotations. Since you're going to be generating faces, you won't need the annotations. You can view the first number of examples by changing show_n_images.

show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'img_align_celeba/*.jpg'))[:show_n_images], 28, 28, 'RGB')
pyplot.imshow(helper.images_square_grid(mnist_images, 'RGB'))

<matplotlib.image.AxesImage at 0x24985b719e8>

Preprocess the Data¶

Since the project's main focus is on building the GANs, we'll preprocess the data for you. The values of the MNIST and CelebA dataset will be in the range of -0.5 to 0.5 of 28x28 dimensional images. The CelebA images will be cropped to remove parts of the image that don't include a face, then resized down to 28x28.

The MNIST images are black and white images with a single color channel while the CelebA images have 3 color channels (RGB color channel).

Build the Neural Network¶

You'll build the components necessary to build a GANs by implementing the following functions below:

model_inputs
discriminator
generator
model_loss
model_opt
train

Check the Version of TensorFlow and Access to GPU¶

This will check to make sure you have the correct version of TensorFlow and access to a GPU

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
from distutils.version import LooseVersion
import warnings
import tensorflow as tf

# Check TensorFlow Version
assert LooseVersion(tf.__version__) >= LooseVersion('1.0'), 'Please use TensorFlow version 1.0 or newer.  You are using {}'.format(tf.__version__)
print('TensorFlow Version: {}'.format(tf.__version__))

# Check for a GPU
if not tf.test.gpu_device_name():
    warnings.warn('No GPU found. Please use a GPU to train your neural network.')
else:
    print('Default GPU Device: {}'.format(tf.test.gpu_device_name()))

TensorFlow Version: 1.4.0
Default GPU Device: /device:GPU:0

Input¶

Implement the model_inputs function to create TF Placeholders for the Neural Network. It should create the following placeholders:

Real input images placeholder with rank 4 using image_width, image_height, and image_channels.
Z input placeholder with rank 2 using z_dim.
Learning rate placeholder with rank 0.

Return the placeholders in the following the tuple (tensor of real input images, tensor of z data)

import problem_unittests as tests

def model_inputs(image_width, image_height, image_channels, z_dim):
    """
    Create the model inputs
    :param image_width: The input image width
    :param image_height: The input image height
    :param image_channels: The number of image channels
    :param z_dim: The dimension of Z
    :return: Tuple of (tensor of real input images, tensor of z data, learning rate)
    """
    # TODO: Implement Function

    real_inputs = tf.placeholder(tf.float32, (None, image_width, image_height, image_channels),
                                 'real_inputs')
    inputs_z = tf.placeholder(tf.float32, (None, z_dim), name='input_z')  
    learning_rate = tf.placeholder(tf.float32, name='learning_rate')
    
    return real_inputs, inputs_z, learning_rate


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_inputs(model_inputs)

Tests Passed

Discriminator¶

Implement discriminator to create a discriminator neural network that discriminates on images. This function should be able to reuse the variables in the neural network. Use tf.variable_scope with a scope name of "discriminator" to allow the variables to be reused. The function should return a tuple of (tensor output of the discriminator, tensor logits of the discriminator).

def discriminator(images, reuse=False):
    """
    Create the discriminator network
    :param images: Tensor of input image(s)
    :param reuse: Boolean if the weights should be reused
    :return: Tuple of (tensor output of the discriminator, tensor logits of the discriminator)
    """
    # TODO: Implement Function
    alpha = 0.2
    
    with tf.variable_scope('discriminator', reuse=reuse):
        # Input layer is 32x32x3
        x1 = tf.layers.conv2d(images, 64, 5, strides=2, padding='same')
        relu1 = tf.maximum(alpha * x1, x1)
        # 16x16x64
        
        x2 = tf.layers.conv2d(relu1, 128, 5, strides=2, padding='same')
        bn2 = tf.layers.batch_normalization(x2, training=True)
        relu2 = tf.maximum(alpha * bn2, bn2)
        # 8x8x128
        
        x3 = tf.layers.conv2d(relu2, 256, 5, strides=2, padding='same')
        bn3 = tf.layers.batch_normalization(x3, training=True)
        relu3 = tf.maximum(alpha * bn3, bn3)
        # 4x4x256

        # Flatten it
        flat = tf.reshape(relu3, (-1, 4*4*256))
        logits = tf.layers.dense(flat, 1)
        out = tf.sigmoid(logits)
        
        return out, logits
"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_discriminator(discriminator, tf)

Tests Passed

Generator¶

Implement generator to generate an image using z. This function should be able to reuse the variables in the neural network. Use tf.variable_scope with a scope name of "generator" to allow the variables to be reused. The function should return the generated 28 x 28 x out_channel_dim images.

def generator(z, out_channel_dim, is_train=True):
    """
    Create the generator network
    :param z: Input z
    :param out_channel_dim: The number of channels in the output image
    :param is_train: Boolean if generator is being used for training
    :return: The tensor output of the generator
    """
    # TODO: Implement Function
    
    reuse = not is_train
    
    with tf.variable_scope('generator', reuse=reuse):
        alpha = 0.2
        
        x1 = tf.layers.dense(z, 2*2*512)
        x1 = tf.reshape(x1, (-1, 2, 2, 512))
        x1 = tf.layers.batch_normalization(x1, training=is_train)
        x1 = tf.maximum(alpha * x1, x1)
        
        x2 = tf.layers.conv2d_transpose(x1, 256, 5, 2, 'valid')
        x2 = tf.layers.batch_normalization(x2, training=is_train)
        x2 = tf.maximum(alpha * x2, x2)
        
        x3 = tf.layers.conv2d_transpose(x2, 128, 5, 2, 'same')
        x3 = tf.layers.batch_normalization(x3, training=is_train)
        x3 = tf.maximum(alpha * x3, x3)
        
        logits = tf.layers.conv2d_transpose(x3, out_channel_dim, 5, 2, 'same')
        out = tf.tanh(logits)
        
        return out

"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_generator(generator, tf)

Tests Passed

Loss¶

Implement model_loss to build the GANs for training and calculate the loss. The function should return a tuple of (discriminator loss, generator loss). Use the following functions you implemented:

discriminator(images, reuse=False)
generator(z, out_channel_dim, is_train=True)

def model_loss(input_real, input_z, out_channel_dim):
    """
    Get the loss for the discriminator and generator
    :param input_real: Images from the real dataset
    :param input_z: Z input
    :param out_channel_dim: The number of channels in the output image
    :return: A tuple of (discriminator loss, generator loss)
    """
    # TODO: Implement Function
    
    g_model = generator(input_z, out_channel_dim)
    d_model_real, d_logits_real = discriminator(input_real)
    d_model_fake, d_logits_fake = discriminator(g_model, reuse=True)

    # Label Smoothing. Salimans et. al. 2016 recommend to only smooth the positive labels
    smooth = 0.1

    d_loss_real = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_real, labels=tf.ones_like(d_model_real) * (1 - smooth)))
    d_loss_fake = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, labels=tf.zeros_like(d_model_fake)))
    g_loss = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, labels=tf.ones_like(d_model_fake)))

    d_loss = d_loss_real + d_loss_fake

    return d_loss, g_loss


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_loss(model_loss)

Tests Passed

Optimization¶

Implement model_opt to create the optimization operations for the GANs. Use tf.trainable_variables to get all the trainable variables. Filter the variables with names that are in the discriminator and generator scope names. The function should return a tuple of (discriminator training operation, generator training operation).

def model_opt(d_loss, g_loss, learning_rate, beta1):
    """
    Get optimization operations
    :param d_loss: Discriminator loss Tensor
    :param g_loss: Generator loss Tensor
    :param learning_rate: Learning Rate Placeholder
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :return: A tuple of (discriminator training operation, generator training operation)
    """
    # TODO: Implement Function
    t_vars = tf.trainable_variables()
    d_vars = [var for var in t_vars if var.name.startswith('discriminator')]
    g_vars = [var for var in t_vars if var.name.startswith('generator')]
        
    ops = tf.get_collection(tf.GraphKeys.UPDATE_OPS)
    g_updates = [opt for opt in ops if opt.name.startswith('generator')]
    
    with tf.control_dependencies(g_updates):
        g_train_opt = tf.train.AdamOptimizer(
            learning_rate, beta1).minimize(g_loss, var_list=g_vars)
        d_train_opt = tf.train.AdamOptimizer(
            learning_rate, beta1=beta1).minimize(d_loss, var_list=d_vars)
        
    return d_train_opt, g_train_opt


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_opt(model_opt, tf)

Tests Passed

Neural Network Training¶

Show Output¶

Use this function to show the current output of the generator during training. It will help you determine how well the GANs is training.

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import numpy as np

def show_generator_output(sess, n_images, input_z, out_channel_dim, image_mode):
    """
    Show example output for the generator
    :param sess: TensorFlow session
    :param n_images: Number of Images to display
    :param input_z: Input Z Tensor
    :param out_channel_dim: The number of channels in the output image
    :param image_mode: The mode to use for images ("RGB" or "L")
    """
    
    
    
    
    cmap = None if image_mode == 'RGB' else 'gray'
    z_dim = input_z.get_shape().as_list()[-1]
    example_z = np.random.uniform(-1, 1, size=[n_images, z_dim])

    samples = sess.run(
        generator(input_z, out_channel_dim, False),
        feed_dict={input_z: example_z})

    images_grid = helper.images_square_grid(samples, image_mode)
    
    # image was converted to np.array, otherwise don't work for gray scale images
    #pyplot.imshow(images_grid, cmap=cmap)
    
    # This solution was given to me by 'Itai Dror' in the Slack Chat
    
    pyplot.imshow(np.array(helper.images_square_grid(samples, image_mode)), cmap=cmap)
    
    
    pyplot.show()

Train¶

Implement train to build and train the GANs. Use the following functions you implemented:

model_inputs(image_width, image_height, image_channels, z_dim)
model_loss(input_real, input_z, out_channel_dim)
model_opt(d_loss, g_loss, learning_rate, beta1)

Use the show_generator_output to show generator output while you train. Running show_generator_output for every batch will drastically increase training time and increase the size of the notebook. It's recommended to print the generator output every 100 batches.

def train(epoch_count, batch_size, z_dim, learning_rate, beta1, get_batches, data_shape, data_image_mode):
    """
    Train the GAN
    :param epoch_count: Number of epochs
    :param batch_size: Batch Size
    :param z_dim: Z dimension
    :param learning_rate: Learning Rate
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :param get_batches: Function to get batches
    :param data_shape: Shape of the data
    :param data_image_mode: The image mode to use for images ("RGB" or "L")
    """
    # TODO: Build Model
    _, img_width, img_height, img_channels = data_shape
    
    real_input, z_input, lr = model_inputs(
        img_width, img_height, img_channels, z_dim)
    
    d_loss, g_loss = model_loss(real_input, z_input, img_channels)
    d_opt, g_opt = model_opt(d_loss, g_loss, learning_rate, beta1)
    
    steps = 0
    print_every = 10
    show_every = 100
    losses = []
    n_images = 25
    
    
    with tf.Session() as sess:
        sess.run(tf.global_variables_initializer())
        for epoch_i in range(epoch_count):
            for batch_images in get_batches(batch_size):
             # TODO: Train Model
                steps += 1
                batch_images *= 2.0
                z_sample = np.random.uniform(-1, 1, (batch_size, z_dim))
                
                _ = sess.run(d_opt, feed_dict={
                        real_input: batch_images, z_input: z_sample, lr: learning_rate})
                _ = sess.run(g_opt, feed_dict={
                        z_input: z_sample, lr: learning_rate})
                
                if steps % print_every == 0:
                    train_loss_d = d_loss.eval({z_input: z_sample, real_input: batch_images})
                    train_loss_g = g_loss.eval({z_input: z_sample})
                    
                    print("Epoch {}/{}...".format(epoch_i+1, epoch_count),
                          "Discriminator Loss: {:.4f}...".format(train_loss_d),
                          "Generator Loss: {:.4f}".format(train_loss_g))
                    losses.append((train_loss_d, train_loss_g))
                
                if steps % show_every == 0:
                    show_generator_output(sess, n_images, z_input, img_channels, data_image_mode)

                    
    #plot
      
    fig, ax = pyplot.subplots()
    for i, (loss_d, g_loss) in enumerate(losses):
#         scale = 200.0 * rand(n)
        ax.scatter(i, loss_d, c='red')
        ax.scatter(i, g_loss, c='blue')
    print("Red: Discriminator Loss, Ble: Generator Loss")
    pyplot.show()

MNIST¶

Test your GANs architecture on MNIST. After 2 epochs, the GANs should be able to generate images that look like handwritten digits. Make sure the loss of the generator is lower than the loss of the discriminator or close to 0.

batch_size = 16
z_dim = 100
learning_rate = 0.0002
beta1 = 0.5


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 2

mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'img_align_celeba/*.jpg'))[:show_n_images], 28, 28, 'RGB')
mnist_dataset = helper.Dataset('mnist', glob(os.path.join(data_dir, 'mnist/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, mnist_dataset.get_batches,
          mnist_dataset.shape, mnist_dataset.image_mode)

Epoch 1/2... Discriminator Loss: 0.7237... Generator Loss: 1.2118
Epoch 1/2... Discriminator Loss: 1.6683... Generator Loss: 0.3767
Epoch 1/2... Discriminator Loss: 0.5189... Generator Loss: 2.4045
Epoch 1/2... Discriminator Loss: 0.4605... Generator Loss: 2.4838
Epoch 1/2... Discriminator Loss: 0.8115... Generator Loss: 1.0498
Epoch 1/2... Discriminator Loss: 0.9422... Generator Loss: 0.8721
Epoch 1/2... Discriminator Loss: 0.7456... Generator Loss: 1.9276
Epoch 1/2... Discriminator Loss: 0.4170... Generator Loss: 7.0585
Epoch 1/2... Discriminator Loss: 0.4816... Generator Loss: 3.2120
Epoch 1/2... Discriminator Loss: 0.4400... Generator Loss: 3.0955

Epoch 1/2... Discriminator Loss: 0.6740... Generator Loss: 1.7621
Epoch 1/2... Discriminator Loss: 0.8440... Generator Loss: 5.6265
Epoch 1/2... Discriminator Loss: 0.8900... Generator Loss: 1.2756
Epoch 1/2... Discriminator Loss: 0.5689... Generator Loss: 4.6694
Epoch 1/2... Discriminator Loss: 0.4452... Generator Loss: 3.1377
Epoch 1/2... Discriminator Loss: 0.6456... Generator Loss: 3.0443
Epoch 1/2... Discriminator Loss: 0.5305... Generator Loss: 2.5497
Epoch 1/2... Discriminator Loss: 0.4017... Generator Loss: 3.3831
Epoch 1/2... Discriminator Loss: 0.4235... Generator Loss: 3.5784
Epoch 1/2... Discriminator Loss: 0.3886... Generator Loss: 3.4951

Epoch 1/2... Discriminator Loss: 0.4470... Generator Loss: 2.4830
Epoch 1/2... Discriminator Loss: 0.4255... Generator Loss: 2.7618
Epoch 1/2... Discriminator Loss: 0.4084... Generator Loss: 3.3357
Epoch 1/2... Discriminator Loss: 0.3861... Generator Loss: 3.5096
Epoch 1/2... Discriminator Loss: 0.4030... Generator Loss: 3.5160
Epoch 1/2... Discriminator Loss: 0.3843... Generator Loss: 3.5867
Epoch 1/2... Discriminator Loss: 0.3733... Generator Loss: 3.7320
Epoch 1/2... Discriminator Loss: 0.4782... Generator Loss: 2.5373
Epoch 1/2... Discriminator Loss: 0.4552... Generator Loss: 3.4076
Epoch 1/2... Discriminator Loss: 0.4236... Generator Loss: 3.5211

Epoch 1/2... Discriminator Loss: 1.0010... Generator Loss: 8.2480
Epoch 1/2... Discriminator Loss: 0.4478... Generator Loss: 2.7508
Epoch 1/2... Discriminator Loss: 0.4794... Generator Loss: 3.4031
Epoch 1/2... Discriminator Loss: 0.5162... Generator Loss: 2.7756
Epoch 1/2... Discriminator Loss: 0.7187... Generator Loss: 1.2564
Epoch 1/2... Discriminator Loss: 0.5146... Generator Loss: 2.2330
Epoch 1/2... Discriminator Loss: 0.4835... Generator Loss: 2.7279
Epoch 1/2... Discriminator Loss: 0.4541... Generator Loss: 2.8404
Epoch 1/2... Discriminator Loss: 0.6196... Generator Loss: 1.7910
Epoch 1/2... Discriminator Loss: 0.4190... Generator Loss: 5.3511

Epoch 1/2... Discriminator Loss: 0.5231... Generator Loss: 2.3384
Epoch 1/2... Discriminator Loss: 0.5314... Generator Loss: 3.3628
Epoch 1/2... Discriminator Loss: 0.6246... Generator Loss: 1.7644
Epoch 1/2... Discriminator Loss: 0.6511... Generator Loss: 1.6439
Epoch 1/2... Discriminator Loss: 0.5159... Generator Loss: 2.3954
Epoch 1/2... Discriminator Loss: 0.5242... Generator Loss: 2.1987
Epoch 1/2... Discriminator Loss: 0.4820... Generator Loss: 2.4765
Epoch 1/2... Discriminator Loss: 0.6253... Generator Loss: 2.1763
Epoch 1/2... Discriminator Loss: 0.7282... Generator Loss: 1.6720
Epoch 1/2... Discriminator Loss: 0.6071... Generator Loss: 2.8879

Epoch 1/2... Discriminator Loss: 0.6132... Generator Loss: 1.8793
Epoch 1/2... Discriminator Loss: 0.7015... Generator Loss: 1.7890
Epoch 1/2... Discriminator Loss: 0.7842... Generator Loss: 4.3030
Epoch 1/2... Discriminator Loss: 0.6694... Generator Loss: 1.5348
Epoch 1/2... Discriminator Loss: 0.5444... Generator Loss: 2.9620
Epoch 1/2... Discriminator Loss: 0.6717... Generator Loss: 3.1639
Epoch 1/2... Discriminator Loss: 0.8072... Generator Loss: 1.4684
Epoch 1/2... Discriminator Loss: 0.5609... Generator Loss: 2.0871
Epoch 1/2... Discriminator Loss: 0.4559... Generator Loss: 3.0913
Epoch 1/2... Discriminator Loss: 0.5567... Generator Loss: 2.0557

Epoch 1/2... Discriminator Loss: 0.4675... Generator Loss: 2.7748
Epoch 1/2... Discriminator Loss: 0.8298... Generator Loss: 3.8198
Epoch 1/2... Discriminator Loss: 0.4942... Generator Loss: 3.0026
Epoch 1/2... Discriminator Loss: 1.0590... Generator Loss: 0.7960
Epoch 1/2... Discriminator Loss: 0.7945... Generator Loss: 2.5137
Epoch 1/2... Discriminator Loss: 0.6025... Generator Loss: 2.9383
Epoch 1/2... Discriminator Loss: 0.7579... Generator Loss: 2.0040
Epoch 1/2... Discriminator Loss: 0.5179... Generator Loss: 2.0334
Epoch 1/2... Discriminator Loss: 0.6136... Generator Loss: 2.1189
Epoch 1/2... Discriminator Loss: 0.7124... Generator Loss: 1.8738

Epoch 1/2... Discriminator Loss: 0.9264... Generator Loss: 1.1289
Epoch 1/2... Discriminator Loss: 0.9640... Generator Loss: 2.0160
Epoch 1/2... Discriminator Loss: 0.7286... Generator Loss: 1.9458
Epoch 1/2... Discriminator Loss: 0.7015... Generator Loss: 2.2119
Epoch 1/2... Discriminator Loss: 1.1041... Generator Loss: 1.2841
Epoch 1/2... Discriminator Loss: 0.7400... Generator Loss: 1.6917
Epoch 1/2... Discriminator Loss: 0.7152... Generator Loss: 2.0376
Epoch 1/2... Discriminator Loss: 0.6037... Generator Loss: 2.4632
Epoch 1/2... Discriminator Loss: 0.4656... Generator Loss: 2.9536
Epoch 1/2... Discriminator Loss: 0.6203... Generator Loss: 2.2571

Epoch 1/2... Discriminator Loss: 0.7043... Generator Loss: 2.0390
Epoch 1/2... Discriminator Loss: 0.5770... Generator Loss: 2.5516
Epoch 1/2... Discriminator Loss: 0.9892... Generator Loss: 1.4154
Epoch 1/2... Discriminator Loss: 0.6999... Generator Loss: 1.8235
Epoch 1/2... Discriminator Loss: 1.0701... Generator Loss: 0.8287
Epoch 1/2... Discriminator Loss: 1.1373... Generator Loss: 3.0875
Epoch 1/2... Discriminator Loss: 1.0161... Generator Loss: 0.8728
Epoch 1/2... Discriminator Loss: 0.9287... Generator Loss: 1.6032
Epoch 1/2... Discriminator Loss: 0.6541... Generator Loss: 1.8716
Epoch 1/2... Discriminator Loss: 1.0927... Generator Loss: 0.7763

Epoch 1/2... Discriminator Loss: 0.9031... Generator Loss: 1.9138
Epoch 1/2... Discriminator Loss: 0.9986... Generator Loss: 1.1277
Epoch 1/2... Discriminator Loss: 0.7106... Generator Loss: 1.5738
Epoch 1/2... Discriminator Loss: 0.7078... Generator Loss: 2.2120
Epoch 1/2... Discriminator Loss: 0.9149... Generator Loss: 1.1927
Epoch 1/2... Discriminator Loss: 0.8229... Generator Loss: 1.2071
Epoch 1/2... Discriminator Loss: 0.7876... Generator Loss: 1.3271
Epoch 1/2... Discriminator Loss: 0.9651... Generator Loss: 1.2101
Epoch 1/2... Discriminator Loss: 0.8340... Generator Loss: 1.2916
Epoch 1/2... Discriminator Loss: 0.7531... Generator Loss: 1.6166

CelebA¶

Run your GANs on CelebA. It will take around 20 minutes on the average GPU to run one epoch. You can run the whole epoch or stop when it starts to generate realistic faces.

%%time
batch_size = 16
z_dim = 100
learning_rate = 0.0004
beta1 = 0.5


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 1

celeba_dataset = helper.Dataset('celeba', glob(os.path.join(data_dir, 'img_align_celeba/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, celeba_dataset.get_batches,
          celeba_dataset.shape, celeba_dataset.image_mode)

Epoch 1/1... Discriminator Loss: 0.7309... Generator Loss: 1.3716
Epoch 1/1... Discriminator Loss: 2.8461... Generator Loss: 0.0935
Epoch 1/1... Discriminator Loss: 2.3512... Generator Loss: 0.1643
Epoch 1/1... Discriminator Loss: 0.4932... Generator Loss: 4.1807
Epoch 1/1... Discriminator Loss: 2.1767... Generator Loss: 0.2037
Epoch 1/1... Discriminator Loss: 0.4477... Generator Loss: 3.5077
Epoch 1/1... Discriminator Loss: 2.0124... Generator Loss: 0.2279
Epoch 1/1... Discriminator Loss: 0.9828... Generator Loss: 0.9711
Epoch 1/1... Discriminator Loss: 0.4794... Generator Loss: 6.5697
Epoch 1/1... Discriminator Loss: 0.7022... Generator Loss: 1.8602

Epoch 1/1... Discriminator Loss: 1.5555... Generator Loss: 8.2803
Epoch 1/1... Discriminator Loss: 1.5256... Generator Loss: 0.5654
Epoch 1/1... Discriminator Loss: 0.8644... Generator Loss: 0.9926
Epoch 1/1... Discriminator Loss: 0.8409... Generator Loss: 3.7744
Epoch 1/1... Discriminator Loss: 0.8590... Generator Loss: 1.1751
Epoch 1/1... Discriminator Loss: 1.1738... Generator Loss: 0.7413
Epoch 1/1... Discriminator Loss: 0.6105... Generator Loss: 4.2648
Epoch 1/1... Discriminator Loss: 1.0142... Generator Loss: 0.8589
Epoch 1/1... Discriminator Loss: 0.6065... Generator Loss: 3.3037
Epoch 1/1... Discriminator Loss: 0.5620... Generator Loss: 2.2098

Epoch 1/1... Discriminator Loss: 0.8728... Generator Loss: 3.3832
Epoch 1/1... Discriminator Loss: 0.8297... Generator Loss: 2.2165
Epoch 1/1... Discriminator Loss: 1.1648... Generator Loss: 0.6418
Epoch 1/1... Discriminator Loss: 0.7864... Generator Loss: 1.7156
Epoch 1/1... Discriminator Loss: 1.2351... Generator Loss: 0.6205
Epoch 1/1... Discriminator Loss: 0.8839... Generator Loss: 1.2668
Epoch 1/1... Discriminator Loss: 1.5548... Generator Loss: 0.6031
Epoch 1/1... Discriminator Loss: 1.0488... Generator Loss: 2.2887
Epoch 1/1... Discriminator Loss: 1.6018... Generator Loss: 0.8860
Epoch 1/1... Discriminator Loss: 0.9003... Generator Loss: 1.6896

Epoch 1/1... Discriminator Loss: 1.2785... Generator Loss: 0.7501
Epoch 1/1... Discriminator Loss: 0.9877... Generator Loss: 1.0029
Epoch 1/1... Discriminator Loss: 2.0155... Generator Loss: 0.3295
Epoch 1/1... Discriminator Loss: 1.1321... Generator Loss: 1.3387
Epoch 1/1... Discriminator Loss: 0.9676... Generator Loss: 1.5814
Epoch 1/1... Discriminator Loss: 1.0653... Generator Loss: 1.0945
Epoch 1/1... Discriminator Loss: 1.0115... Generator Loss: 1.1464
Epoch 1/1... Discriminator Loss: 1.3881... Generator Loss: 1.2127
Epoch 1/1... Discriminator Loss: 1.3057... Generator Loss: 0.9633
Epoch 1/1... Discriminator Loss: 1.3938... Generator Loss: 0.7282

Epoch 1/1... Discriminator Loss: 0.9969... Generator Loss: 1.3175
Epoch 1/1... Discriminator Loss: 1.0821... Generator Loss: 1.7218
Epoch 1/1... Discriminator Loss: 0.9407... Generator Loss: 1.1580
Epoch 1/1... Discriminator Loss: 1.0321... Generator Loss: 0.8892
Epoch 1/1... Discriminator Loss: 1.3075... Generator Loss: 0.8124
Epoch 1/1... Discriminator Loss: 1.7523... Generator Loss: 0.4883
Epoch 1/1... Discriminator Loss: 1.2712... Generator Loss: 0.7915
Epoch 1/1... Discriminator Loss: 1.0568... Generator Loss: 0.8702
Epoch 1/1... Discriminator Loss: 1.2393... Generator Loss: 0.6678
Epoch 1/1... Discriminator Loss: 1.0896... Generator Loss: 1.0032

Epoch 1/1... Discriminator Loss: 1.2441... Generator Loss: 0.9382
Epoch 1/1... Discriminator Loss: 0.9767... Generator Loss: 1.2141
Epoch 1/1... Discriminator Loss: 1.1354... Generator Loss: 0.8147
Epoch 1/1... Discriminator Loss: 0.8407... Generator Loss: 1.3749
Epoch 1/1... Discriminator Loss: 1.1853... Generator Loss: 1.2613
Epoch 1/1... Discriminator Loss: 1.0731... Generator Loss: 1.3675
Epoch 1/1... Discriminator Loss: 1.3228... Generator Loss: 1.0669
Epoch 1/1... Discriminator Loss: 1.3610... Generator Loss: 0.6675
Epoch 1/1... Discriminator Loss: 1.2056... Generator Loss: 1.1885
Epoch 1/1... Discriminator Loss: 1.1233... Generator Loss: 1.3089

Epoch 1/1... Discriminator Loss: 0.9248... Generator Loss: 1.2445
Epoch 1/1... Discriminator Loss: 1.0203... Generator Loss: 1.0375
Epoch 1/1... Discriminator Loss: 1.0781... Generator Loss: 0.9225
Epoch 1/1... Discriminator Loss: 1.2292... Generator Loss: 0.7356
Epoch 1/1... Discriminator Loss: 0.9517... Generator Loss: 1.4331
Epoch 1/1... Discriminator Loss: 0.9047... Generator Loss: 1.2117
Epoch 1/1... Discriminator Loss: 0.8115... Generator Loss: 1.5301
Epoch 1/1... Discriminator Loss: 1.4306... Generator Loss: 0.5571
Epoch 1/1... Discriminator Loss: 0.9989... Generator Loss: 1.1226
Epoch 1/1... Discriminator Loss: 1.4740... Generator Loss: 0.5968

Epoch 1/1... Discriminator Loss: 0.9806... Generator Loss: 1.2866
Epoch 1/1... Discriminator Loss: 1.3027... Generator Loss: 0.6659
Epoch 1/1... Discriminator Loss: 1.0495... Generator Loss: 1.0183
Epoch 1/1... Discriminator Loss: 1.1645... Generator Loss: 0.7280
Epoch 1/1... Discriminator Loss: 1.3802... Generator Loss: 0.8196
Epoch 1/1... Discriminator Loss: 0.8598... Generator Loss: 1.2798
Epoch 1/1... Discriminator Loss: 1.1732... Generator Loss: 0.8462
Epoch 1/1... Discriminator Loss: 1.2011... Generator Loss: 0.9764
Epoch 1/1... Discriminator Loss: 1.1735... Generator Loss: 1.1243
Epoch 1/1... Discriminator Loss: 1.3204... Generator Loss: 0.6222

Epoch 1/1... Discriminator Loss: 0.9115... Generator Loss: 0.9820
Epoch 1/1... Discriminator Loss: 2.0549... Generator Loss: 0.2683
Epoch 1/1... Discriminator Loss: 0.9526... Generator Loss: 1.0113
Epoch 1/1... Discriminator Loss: 0.9657... Generator Loss: 0.9552
Epoch 1/1... Discriminator Loss: 0.9188... Generator Loss: 1.5391
Epoch 1/1... Discriminator Loss: 1.0944... Generator Loss: 1.1410
Epoch 1/1... Discriminator Loss: 1.0535... Generator Loss: 1.1247
Epoch 1/1... Discriminator Loss: 0.9183... Generator Loss: 1.4498
Epoch 1/1... Discriminator Loss: 0.9771... Generator Loss: 0.8899
Epoch 1/1... Discriminator Loss: 0.9876... Generator Loss: 0.9515

Epoch 1/1... Discriminator Loss: 1.3336... Generator Loss: 0.7512
Epoch 1/1... Discriminator Loss: 0.9297... Generator Loss: 1.2032
Epoch 1/1... Discriminator Loss: 1.1352... Generator Loss: 1.2566
Epoch 1/1... Discriminator Loss: 1.3872... Generator Loss: 0.5523
Epoch 1/1... Discriminator Loss: 1.0349... Generator Loss: 1.0531
Epoch 1/1... Discriminator Loss: 1.1385... Generator Loss: 0.7346
Epoch 1/1... Discriminator Loss: 0.7884... Generator Loss: 1.5097
Epoch 1/1... Discriminator Loss: 0.9322... Generator Loss: 2.1437
Epoch 1/1... Discriminator Loss: 0.9015... Generator Loss: 1.3934
Epoch 1/1... Discriminator Loss: 1.2218... Generator Loss: 0.8187

Submitting This Project¶

When submitting this project, make sure to run all the cells before saving the notebook. Save the notebook file as "dlnd_face_generation.ipynb" and save it as a HTML file under "File" -> "Download as". Include the "helper.py" and "problem_unittests.py" files in your submission.